Skip to content

biodbs 0.4.1

Released: 2026-07-18

0.4.1 fixes SILVA discovery and download integrity after SILVA migrated its website, and adds fetch-only access to the exact HOMD, MOMD, and NCBI artifacts used by FOMC-inspired oral-microbiome pipelines.

SILVA fixes

  • Fixed SILVA file and classifier downloads. SILVA moved to a CMS in which the current-release/ and archive/ paths return HTML browse pages, so downloads were silently saving those pages instead of the requested file. Files are now fetched from SILVA's current file host, fileadmin/silva_databases/current/, and a download that receives an HTML page raises a clear error instead of writing it to disk.
  • Fixed SILVA listings. list_current_files and list_archive_releases returned nothing because SILVA's CMS uses root-relative links; they now return both sub-directories and downloadable file leaves (with is_dir), so classifier filenames and their .md5 sidecars are discoverable.
  • download_classifier now takes the nested path below the classifier directory (SILVA nests classifiers by release and marker, e.g. 2025.7/taxonomic-weights/SILVA_138.2_Ref_NR99_taxonomic-weight_human-oral.qza) and verifies the download against SILVA's published .md5 by default.

New fetch capabilities

  • Shared atomic downloader. Large downloads stream to a temporary file and are moved into place only after the transfer — and any published MD5 — succeeds, so an interrupted download is never cached as a valid file.
  • HOMD and MOMD 16S. list_16s_refseq, download_16s_refseq, and the new download_16s_taxonomy are version- and source-aware (version="15.22", source="homd" or "momd"), selecting the unaligned .fasta and .qiime.taxonomy for a pinned release. MOMD is served from its own momd.org host.
  • NCBI static archives. download_blast_database (e.g. 16S_ribosomal_RNA) and download_taxdump (new_taxdump) fetch NCBI's preformatted archives, MD5-verified against their published sidecars.
  • FOMC Reference Sources documents the end-to-end fetch-only recipe. biodbs downloads and integrity-checks the source artifacts; extraction, taxonomy conversion, reference combination, and the FOMC assignment algorithm remain in the downstream pipeline.

CI Reliability

Added live contract tests for SILVA leaf/MD5 discovery, HOMD/MOMD listings, and NCBI archive reachability, with homd and ncbi added to the CI integration matrix (using HEAD requests and listings so large archives are never downloaded in CI).

Upgrade

pip install --upgrade biodbs==0.4.1

No API migration is required for users upgrading from 0.4.0. If you call download_classifier, pass the classifier's nested path (below the QIIME2/DADA2/... directory) as filename; verification is on by default (verify=False to skip).